First, let me say that I am using Outlook 2016 on a Mac book Pro with OS X 10.10.5. I have trained spamsieve manually as explained in the manual. I am having problems with both very creative ways for “spamvertisers” to avoid spamsieve’s filters, and, an apparent error in spamsieve’s functionality.
Spam originators are employing Random user names and domain names in their addresses which renders spamsieve’s Address filtering virtually ineffective.
This is what a small portion of the rules looked like this morning.
Note that the random username and Random portion of the domain makes the address totally unique. All by itself, this would make the address filter useless the filter was made less “choosy”. Note that the random address parts have been removed, the criteria is changed to “contains” and duplicate rules for the same generic domain have been disabled (then later removed to keep the number of rules lower.) It seems to me that this should make it difficult for advertisers to “slip through” multiple spams on the same “generisized” domain. This seems to be largely defeated by the spam–generators’ changing the generic part of the domain multiple times a day
My attempt to help with at least from a portion of the spam onslaught was to genericize the Name portion of the message as was done with the domain portion. Additionally, I notice that almost every message I got whose domain ended in “.info” was spam, so I’ve made a rule to kill anything coming in with that ending.
With all these “clever” rules in effect, still no filtering occurs whatsoever (zero hits on ALL 350+filters).
I have checked both the white list and the blocked list and removed a few mistakes. The Bayesian classifier doesn’t seem to work since there is a huge similarity/overlap between many spam messages.
I would greatly appreciate anyone’s suggestions as to what I have done wrong or what I can do to make this work. Thanks.